This is the first in what will hopefully be a series of posts by new underneathEM.com recruit Dr Kate Field, an Emergency Physician based at Calvary Hospital in Hobart. In this episode, Kate takes a look at the dogma and assumptions underlying the most commonly (un)prescribed drug delivered in our Emergency Departments.

Subsequent posts will more closely examine the evidence for O2 use in cardiac chest pain, COPD/asthma, pneumonia, sepsis and traumatic brain injury (TBI).

So let’s inhale deeply, fill our FRC with fresh air, and see where Kate takes us:

Oxygen 1: friend or foe?


I became interested in this question a couple of years ago when I was informed by an RN in a regional Australian hospital that they “aren’t allowed” to give high-flow O2 to patients (with normal pulse-oximetry readings) presenting with cardiac-sounding chest pain “because free radicals make myocardial damage worse”.

At the time, as a very fresh consultant/attending Emergency Physician, I felt like I may have missed something important and felt somewhat chastised by my nursing colleagues. I decided to investigate the evidence for O2 therapy in a wide range of conditions.

Traditionally, we were taught that administration of high-flow O2 is standard practice (think back to EMST, APLS and ATLS teaching) for all patients who are critically unwell, and definitely for all cardiac patients. The exception that proved the rule was the chronic CO2-retainer with COPD, but even then, if they were severely hypoxic, they received life-saving O2 (high-flow if required), given the oft-quoted axiom “hypoxia kills quickly, hypercapnia kills slowly”.

But could we have got it wrong? Have we been inadvertently harming our patients?

Some of the questions arising in my quest include:

  • What is ideal PaO2? What is the definition of hyperoxia?
  • What does hyperoxia and free radical generation really do, other than cause wrinkles and help sell expensive face-creams to aging women and Sydney men?
  • Does ideal PaO2 vary in individuals?
  • Is there individual variation depending on biological or pathological processes that are occurring; what is that patient’s ideal PaO2 considering the disease state they are in now?

What I discovered is that there is very limited evidence in the literature.

I was able to establish that “normal” PaO2 on room air (FiO2 = 0.21) at sea level is considered to be 75-100 mmHg. How this translates into individual variation throughout the course of a lifetime varies on patient factors and also disease factors (e.g. smoking, auto-immune disorders, occupational exposures, etc.). Expert consensus suggests that as long as your PaO2 is > 50 mmHg, you’ll be OK. Looking at the Hb-O2 dissociation curve, this roughly correlates with a SpO2 of 85%.

Click here for a refresher on respiratory and Ophysiology.

Individuals will have times during disease burden where they actually have increased O2 consumption. Examples include:

  • Fever: with a 1oC temperature rise, there is a 7% increase in O2 consumption
  • Seizing patients (even if given muscle relaxants): still exhibit a 300-400% increase in O2 consumption in their brain.
  • Septic patients: demonstrate 200% increase in O2 consumption.
  • In exertion/exercise (think seizure, rigors, restlessness, increased work of breathing) your O2 consumption is also going to increase.

How does this information translate to our practice?

The effects of hyperoxia are more variable than just simple free radical production: Hyperoxia can directly affect blood vessels, causing vasoconstriction. At the level of the pulmonary vasculature, this can worsen V/Q mismatch. Vasoconstriction of coronary arteries and cerebral arteries probably isn’t really a good thing either!

How do free radicals actually cause harm? Free radicals are generally mopped up by nitric oxide (NO – the chemical that vasodilates vessels in vivo). They have direct cellular effects, including harming DNA, and can even cause apoptosis of cells.

What is the EVIDENCE for the role of free radicals?

I found this article, which I have to include:

The authors comment that:

“The use of oxygen in “severe angina pectoris” was first described in 1900 by Steele. Clinical improvement after oxygen inhalation in four patients with acute myocardial infarction was reported by Levy and Barach in 1930. Since then the use of oxygen in myocardial infarction has been advised in most standard medical texts.”

(As an aside, I am amazed that “clinical improvement” in only 4 patients in 1930 became standard of care. What else are we doing that has such profound evidence?)

The 1976 BMJ study was a double-blinded RCT which randomized 200 consecutive patients thought to have had MI to treatment with O2 or air, administered via medium concentration mask for the first 24 hours of hospital care. 43 patients were excluded post-hoc when diagnosis of MI was revoked. None received current standard MI treatment (aspirin, thrombolysis, PCI), so the applicability to current practice is somewhat debatable.

There was a mortality rate of 11% in the O2 group vs. 4% in the air group, but this did not achieve statistical significance.

They conclude that the results are suggestive that O2 administration has a “deleterious effect”, and that administration of O2 in patients with an uncomplicated MI does not have any benefit.

[This went unnoticed until unearthed in a systematic review (by Wijesinghe et al. in HEART 2009) of the…wait for it…two, yes, TWO relevant papers in the literature on the subject!

ILCOR and ARC ran with this in 2010-11, and an RN in a regional hospital quoted it back at Kate shortly after – Domhnall/Ed.]

Cementing our move into the 21st century, 2012 brought this paper:

The study:

  • This was a randomised, controlled trial with n = 136.
  • All patients presented with an uncomplicated STEMI (i.e., no cardiogenic shock or “marked hypoxia”).
  • The 2 groups were randomised to receive either “standard of care” (6L/min via medium concentration mask) or titrated O2 aiming for SpO2 93-96%.
  • The primary end-points were 30-day mortality, and infarct size (determined by the Troponin T level at 72 hours)
  • They also utilized MRI at 4-6 weeks for a subset of patients to assess infarct size as a secondary end-point
  • The result was that there was no significant difference between high-concentration O2 vs. titrated O2 in regard to Troponin T concentration, infarct mass or percent infarct mass.

Problems with the study:

  • Small study with wide confidence intervals – the study should probably be repeated with a much larger sample size to truly determine if there is harm/benefit associated with supplemental O2
  • Unblinded study – but, primary end-points were objective measures, so shouldn’t alter the outcomes
  • Pre-hospital O2 therapy (received PRIOR to enrollment in the study) – patients averaged 62 min of pre-hospital O2. Does this have an effect? It certainly muddies the waters for me…
  • There was a significant difference in the two groups with respect to the territories infarcted: more in the high-flow group had an inferior or posterior MI, compared with those in the titrated group with an anterior MI. Those with an anterior MI tended to have higher peak troponins; however this did not lead to a statistically significant difference in the groups

Finally Nikolaou et al. in an opinion-piece in the Hellenic Journal of Cardiology, published in 2012 (53: 329-330) argue that:

“inhalation of 100% O2 for 10-15 minutes is associated with a decreased in coronary blood flow by 20-30% through constriction of the micro vascular resistance vessels”

They hypothesise that this may be due to NO being depleted by mopping up the free radicals generated through hyperoxia. They cite an additional concern that hyperoxia may exacerbate reperfusion injury to the heart (due to increased free radicals).

They conclude that evidence is lacking, but despite this, go on to recommend that

“routine administration of high-flow O2 for ALL acutely ill patients should clearly be abandoned and be replaced with judicious O2 administration guided by pulse oximetry”

I’m rather (un)impressed that they have managed to draw this very concrete conclusion for acutely ill patients (with any aetiology) in a cardiology journal, based on a carefully selected 12 papers, of which 11 are specific to cardiology (the exception was the BTS guidelines for emergency O2 use in adult patients). It really is a bit of a leap, however, this is where I stand:

My conclusions in regards to O2 for ACS:

  • We still don’t know if we cause harm, or benefit, by giving our patients high-flow oxygen when they have myocardial ischaemia.
  • We just don’t have evidence to say that what we have considered as standard of care for years (i.e. 6L/min O2 via Hudson Mask) is harming or benefiting patients in this and other groups of acutely ill patients.

Will my practice change?

  • I think I will probably remove more Hudson Masks now and replace with titrated nasal prongs, if required, aiming for SO2 of 95% (why that number, I can’t say, but it’s what I feel comfortable with and it is within “physiological normalcy”).
  • More importantly, it makes it a hell of a lot easier to take a good history, which will probably benefit my patient more.

Specialist Emergency Physician from Ireland currently based in Tasmania, Australia


  1. Thanks for the thorough review!

    My mindset recently has been this:
    1) As you say, it very likely doesn’t cause harm to have an SpO2 of “only” 95% in the above disease states.
    2) Pushing the SaO2 into supra-physiologic ranges may cause a small amount of harm.
    3) Maxing the patient’s sat to 99% gives me the sense that I can’t judge the efficacy of my interventions or the patient’s changing clinical condition. If they’re at 99% on 15 Lpm and their oxygen exchange begins to improve, I have no objective bedside measure. Also, if they’ve got their O2 blasting and start to deteriorate, I lose a bit of sensitivity in using the SpO2 as a marker of patient condition – by the time it starts to drop when the patient’s on high-flow O2, I must really be behind and the cause could have snowballed quite a bit by this point.
    4) Really, once the SpO2’s at 95%, you don’t gain anything in terms of oxygen transport by pushing the SaO2 if you look at the equation for oxygen carrying capacity (unless you live in a hyperbaric chamber, in which case who even needs hemoglobin…), so I can’t imagine where the real expectation of improved outcomes could come from.
    CaO2=(Hb) x 1.38 x SaO2 + (0.003 x PaO2)

    Again, thanks for the great post, I’m def saving it to reference down the line.

    • Thanks for the feedback – I realised when I heard the throwaway “we don’t do that here” line in the regional hospital that there must be a lot of things that I/we do that we don’t give much consideration to. I really like the fact that this has sent me off on a journey of challenging the dogma…. and that the has been such a revolution in our ability to share these journeys with the concept of FOAM.

      I’m interested as to why we all think Sats of 95% is OK, other than your reasoning in point 4, and our aim…. have you had any more thought on this? Is there any benefit over sats of 93%, or even 97%?

  2. Interesting read – like you, have moved to titrated SpO2 ~95% with nasal specs in past two years

    I am fascinated though – story implies relay of this concept from small regional Hospital RN to new FACEM…and now talking about it in FOAMed blog

    …whatever the ‘right thing to do it’, it is fascinating how change can be slow in medicine and ‘what seems intuitively correct’ may need to be questionned…

  3. A lovely look at how we have got to this remarkably uncertain point.
    What next, perhaps we’ll discover analgesia slows healing? I hope not…
    The lesson is we should always continue to question, and search not just for an answer, but for the right one.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.